Medical Dictation & Transcription Software Compared (2026)

If you are shopping for medical dictation or transcription software, the honest answer is that the best tool depends on one decision you probably have not made yet: do you want to keep dictating, or do you want the note to write itself from the visit. This page compares the named products clinicians actually evaluate, Dragon Copilot, Freed, Sunoh, Suki, and Nabla, with their real strengths, their published prices where they have one, and the workflow tradeoff underneath all of them.

A quick orientation, because the category is muddy. "Dictation software" used to mean push-to-talk: you speak, it types, you still build the note. Most of these tools have since bolted on ambient capture, where the software listens to the whole visit and drafts the note with no dictation at all. So the real choice is not Dragon versus Freed. It is dictation versus ambient. Get that right and the shortlist sorts itself out.

Key takeaways

• The market splits into two workflows: front-end dictation, where you still narrate the note, and ambient capture, where the visit conversation becomes the note.
• Dragon Copilot is the deepest enterprise pick (Epic-native, dictation plus ambient) but publishes no list price; you buy through Microsoft or partners.
• Freed ($39 to $119/mo) and Sunoh ($149/mo promo) are the transparent value picks for solo and small practices; Suki and Nabla publish no pricing at all.
• A 2023 JAMA Network Open study found primary-care physicians log a median 36.2 minutes of EHR time per 30-minute visit, the burden any of these tools is trying to cut.
• AI Scribe by Patient Square competes on the ambient workflow plus a flat, published $89/clinician/month, unlimited notes.

Dictation versus ambient: the choice underneath the shopping

Start here, because it changes which column of the comparison you should care about.

Front-end dictation is the older model. You speak the note out loud and the software transcribes it in real time. The work is faster than typing, but you are still the author composing from memory after the patient leaves. Dragon Medical One built its reputation on doing this well, and the engine is now part of Dragon Copilot.

Ambient capture is the newer model. The software listens to the actual conversation during the visit and drafts a structured note from what was said. There is no dictation step. You see the patient normally and review a draft a couple of minutes later. The narration work disappears instead of getting faster.

Here is the part that trips up evaluations: nearly every tool in this comparison now offers both modes. Dragon Copilot does. Nabla does. Suki does. So comparing them feature-by-feature on a spreadsheet misses the real question, which is which mode you actually want to live in. We break that tradeoff down in detail in our ambient documentation versus dictation guide. If you already know you want a human-in-the-loop typing service rather than software, our AI scribe versus transcription service comparison covers that fork instead.

The rest of this page assumes you are weighing software, and it concedes upfront that several of these products are genuinely good at what they do.

The named products, and what each one is actually good at

No strawmen here. Each of these has a real reason clinicians pick it.

Dragon Copilot (Microsoft / Nuance). The most established name in clinical voice, full stop. It unifies the Dragon Medical One dictation engine with DAX ambient capture in one product, and it is built around Epic: physician documentation in Hyperdrive, nurse flowsheets, radiology via PowerScribe One, write-back through Epic FHIR APIs and the orders module. If your organization runs Epic and wants both dictation and ambient from one vendor with deep EHR write-back, nothing else matches its reach. The catch: there is no published subscription price. You buy through Microsoft reps or certified partners, and consumption runs around $0.01 per unit with an AI-assisted session costing roughly $0.25, per Microsoft's May 2026 licensing guidance. Our Dragon Medical One cost breakdown digs into that pricing model.

Freed. The clearest value play for solo and small practices. Pricing is published and honest: Starter at $39/month (capped at 40 notes), Core at $79/month unlimited, Premier at $119/month adding EHR push and ICD-10 coding. It pushes notes into browser-based EHRs through a Chrome extension and has real scale behind it. If you want a transparent price and a fast setup, Freed earns its shortlist spot.

Sunoh.ai. Distributed by eClinicalWorks, which matters if you are an eCW practice, because the integration is native via healow. Published price is $149 per user per month, promotional, marked down from $199. Ambient capture, draft SOAP, edit before EHR import. A solid pick if you are inside the eCW ecosystem.

Suki. Strong enterprise reach and a polished assistant that does ambient documentation plus dictation and voice commands, with deep bidirectional integrations into Epic, Oracle Health, athenahealth, and MEDITECH. The tradeoff for a shopper: no public pricing, sales-led only. You will be talking to a rep before you see a number.

Nabla. A credible ambient product with both conversation capture and a dedicated dictation mode, real published evidence behind it, and wide EHR coverage. Like Suki, it is sales-led: the pricing page returns a 404, and the fee scales with user count by undisclosed amount. Good tool, opaque price.

The honest summary: if you want depth and you live in Epic, Dragon Copilot. If you want a transparent self-serve price, Freed or Sunoh. If you are fine with a sales call, Suki and Nabla are real. What none of them lead with is a flat, published price for an ambient-first workflow, which is the gap the next section is about.

Medical dictation and transcription software compared

This table is scoped to the things that actually differ across tools and that a clinician can verify. Pricing is per clinician per month, US, as published by each vendor or marked as sales-led where no price exists.

Two rows on that table are worth pausing on, because they are where the products diverge most and where the marketing language gets vaguest.

What happens to the audio

The visit recording is the most sensitive artifact in the whole workflow. It holds everything that did not make the note, the aside about a mental health struggle, the substance use mention, the thing said off the record. So "what happens to the audio, and when is it gone" is the first question to ask any vendor, not the last.

Most tools answer this in their terms rather than on the homepage, and the answers vary: some retain audio for days or weeks for model improvement or quality review, some let you opt out, some are deliberately vague. None of those is automatically disqualifying, but you should get a one-sentence answer with a specific timeline before you sign.

AI Scribe by Patient Square processes visit audio in memory and discards it once the note is drafted. No audio archive exists on our side or the practice's. What survives is the note you reviewed and signed. The full posture is on the security page.

Published price versus "talk to sales"

Of the named ambient tools, only Freed and Sunoh put a real self-serve number in front of you. Dragon Copilot routes you through Microsoft or a partner. Suki and Nabla publish nothing; Nabla actively removed its pricing page. That is a legitimate enterprise sales motion, but for a clinician sizing a single-provider or small-group decision, an opaque price is a cost in itself, the cost of a sales cycle and the uncertainty of where you will land.

AI Scribe by Patient Square is a flat $89 per clinician per month in the US, unlimited notes, published openly on the pricing page. No call required to see the number.

What AI Scribe by Patient Square adds, honestly stated

This is the part where a comparison page is supposed to claim a clean sweep. It will not, because two of these competitors are genuinely deeper than us in specific ways: Dragon Copilot's Epic write-back is unmatched, and Suki's enterprise integration breadth is real. If deep bidirectional EHR integration is your hard requirement, those tools are built for it and we are not.

What we are built for is the ambient workflow at a flat, honest price, with two things bundled in that the dictation-first tools do not lead with.

AI Scribe by Patient Square is an ambient AI medical scribe that listens during the visit and hands back a structured SOAP note, ICD-10 suggestions, and a prescription draft, ready to review and sign about two minutes after the visit. The prescription draft and the ICD-10 suggestions come with the note, not as an upsell tier.

The prescription draft also runs through a deterministic Rx safety screen: drug-interaction, renal, and pregnancy checks, re-screened at sign time, with a hard block and override-with-attestation on the riskiest combinations. That is not a language model guessing about safety; it is a deterministic check on the draft. None of the dictation tools in this comparison ship that. We go deeper on it in our prescription draft safety post.

And it works alongside whatever EHR you already use rather than integrating with it. You review the note and move it into your chart. That keeps setup to about a day. The tradeoff is honest: you do not get one-click write-back into Epic the way Dragon does. For an EHR-agnostic practice, that is usually the better deal; for a deep-Epic shop, it may not be. Pick accordingly.

Does the time savings justify any of this

Worth grounding in the evidence, because the category oversells and the data is more measured than the pitches.

A 2023 JAMA Network Open study found primary-care physicians log a median 36.2 minutes of EHR time per 30-minute visit. The documentation now outlasts the appointment, which is the burden every tool here is trying to cut.

A 2026 JAMA study across five health systems and more than 1,800 clinicians found ambient tools saved about 16 minutes of documentation time per 8-hour shift and about 13 fewer EHR minutes, versus matched controls. The paper's own framing was "modest reductions," which is the honest read. The objective time savings are real but moderate; some studies find the wellbeing gains larger than the raw minutes.

So do not buy on a vendor's "save two hours a day" headline. Buy on what the tool does to your actual note quality and your actual review time, measured on your actual patients. A draft you sign in 90 seconds beats one you spend five minutes fixing, regardless of the sticker.

How to run the evaluation that actually decides it

Every demo runs on a clean exam room and a cooperative patient. Your Thursday clinic is neither.

Pull your five hardest visit types from last week: the patient who talks in circles, the one with three complaints, the one whose family member answers most questions, the one with an accent your autocomplete already mangles, the one whose visit went somewhere unexpected. Run those through any tool on your shortlist and read every draft closely.

Check the things that separate a real note from a plausible one. Does the assessment match your clinical reasoning, or the most likely diagnosis given the chief complaint? Are medication names, doses, and frequencies right? Is anything in the note that was not actually said? A tool that invents a denial of symptoms is worse than no tool.

Then do the price arithmetic on your real volume, not a vendor's example. A flat fee is predictable; a per-session or per-line model moves with your busiest days, which are exactly the days you most need help.

AI Scribe by Patient Square includes a 7-day free trial, enough for one real clinic week. The pricing page shows the full $89/month rate with no sales call. Run it against whatever else is on your shortlist, read the drafts, and you will have your answer by Friday.